sign in sign up
<<
Home , L chondrite

& Planetary Science 38, Nr 10, 1507–1520 (2003) Abstract available online at http://meteoritics.org

Spade: An H impact-melt that experienced post-shock annealing

Alan E. RUBIN 1* and Rhian H. JONES 2

1Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California 90095–1567, USA 2Institute of Meteoritics, Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, New Mexico 87131, USA *Corresponding author. E-mail: [email protected]

(Received 26 January 2003; revision accepted 19 October 2003)

Abstract–The low modal abundances of relict (1.8 vol%) and of coarse (i.e., ³200 mm- size) isolated mafic silicate grains (1.8 vol%) in Spade relative to mean H6 (11.4 and 9.8 vol%, respectively) show Spade to be a rock that has experienced a significant degree of melting. Various petrographic features (e.g., -plagioclase assemblages, chromite veinlets, silicate darkening) indicate that melting was caused by shock. Plagioclase was melted during the shock event and flowed so that it partially to completely surrounded nearby mafic silicate grains. During crystallization, plagioclase developed igneous zoning. Low-Ca that crystallized from the melt (or equilibrated with the melt at high temperatures) acquired relatively high amounts of CaO. Metallic Fe-Ni cooled rapidly below the Fe-Ni solvus and transformed into martensite. Subsequent reheating of the rock caused transformation of martensite into abundant duplex . Ambiguities exist in the shock stage assignment of Spade. The extensive silicate darkening, the occurrence of chromite-plagioclase assemblages, and the impact-melted characteristics of Spade are consistent with shock stage S6. Low shock (stage S2) is indicated by the undulose extinction and lack of planar fractures in . This suggests that Spade reached a maximum prior shock level equivalent to stage S6 and then experienced post-shock annealing (probably to stage S1). These events were followed by a less intense impact that produced the undulose extinction in the olivine, characteristic of shock stage S2. Annealing could have occurred if Spade were emplaced near impact melts beneath the crater floor or deposited in close proximity to hot debris within an ejecta blanket. Spade firmly establishes the case for post-shock annealing. This may have been a common process on OC .

INTRODUCTION blanket. The identification of an OC impact-melt breccia that was shocked and annealed would confirm the reality of Many ordinary chondrites (OC) exhibit evidence of complex thermal processing on OC asteroids. Different impact melting. Many contain impact-melt-rock clasts (e.g., components of the could record different episodes Plainview, Tysnes Island, Johnson City) that formed from of the rock’s thermal and shock history. small-scale impacts into the host (Fodor and Keil 1976; Spade appears to be an annealed impact-melt breccia. Wilkening 1978; Rubin et al. 1983). Some OC are impact- The meteorite was found as a single 8.9 kg stone at a location melt (e.g., Rose City, Cat Mountain, Chico) that 8.5 km north and 1.6 km east of the town of Spade in Lamb contain large impact-melted regions adjacent to relict County, Texas. chondritic regions (Mason and Wiik 1966; Kring et al. 1996; Bogard et al. 1995). A few OC are impact-melt rocks (e.g., ANALYTICAL PROCEDURES PAT 91501, Y-74160, Y-790964, Ramsdorf) that have been extensively or completely melted (Mittlefehldt and Lindstrom Polished thin section UNM 1099 (~185 mm 2) of Spade 2001; Takeda et al. 1984; Yamaguchi et al. 1998, 1999); they was studied microscopically in transmitted and reflected light. contain little relict chondritic material. Grain sizes were measured with a petrographic microscope Some OC have been shocked and annealed (e.g., Rubin using a calibrated reticle. Modal abundances were determined 2002, Forthcoming). Annealing may have been caused by microscopically using an automated point counter. Polished burial of these samples in proximity to impact melts beneath slab C359.1 (~2500 mm 2) from the collection of the Institute the floor of an impact crater or near hot debris in an ejecta of Meteoritics at the University of New Mexico was examined

1507 © , 2003. Printed in USA. 1508 A. E. Rubin and R. H. Jones microscopically. The thin section and slab were etched with a This is much higher than the values for L and LL chondrite dilute solution of nitric acid to reveal metallographic details. falls (8.3 and 3.6 wt%, respectively; Jarosewich 1990). Quantitative analyses of olivine, plagioclase, chromite, Mineral compositions also indicate that Spade is an H- metal, and sulfide were performed using the JEOL 733 group chondrite. The mean olivine and low-Ca pyroxene Superprobe electron microprobe at UNM, with operating compositions in Spade (Fa 19.0 and Fs16.0, respectively; Table conditions of 15 kV and a beam current of 20 nA. Silicate 2) are within the H4–6 chondrite ranges (Fa 16–20, Fs14.5–18; standards were used for silicate minerals, and a combination Brearley and Jones 1998). The mean plagioclase composition of metal and sulfide standards was used for metal and sulfide (n = 24) of Ab 79.9Or4.8 (Table 2) is closest to the mean of H mineral analyses. Low-Ca pyroxene and Ca pyroxene were chondrite plagioclase (Ab 81.9Or5.8; V an Schmus and Ribbe analyzed with the JEOL JXA-8200 electron microprobe at 1968) and appreciably less sodic than the mean compositions UCLA, using natural and synthetic standards, an accelerating of plagioclase in L and LL chondrites (i.e., Ab 84.2Or5.6 and voltage of 15 keV, a 15 nA sample current, 20 sec counting Ab85.9Or3.6, respectively). The mean Co content of times, and ZAF corrections. Spade (0.46 ± 0.03 wt%; Table 3) is within the range (0.44–0.51 wt% Co; Rubin 1990) and the kamacite Ni RESULTS content (6.9 wt%; Table 3) is identical to the mean value in H4–6 chondrites (Rubin 1990). Weathering Grade Texture and Mineral Chemistry Thin section and slab examination indicate that ~60% of the silicate grains exhibit significant brown staining. Many The Spade whole-rock does not appear to be brecciated; mafic silicate grains are traversed and surrounded by 0.1 mm- no clasts are visible in the slab. The rock is rich in metallic Fe- thick veins of limonite ( a-FeO[OH]). The majority of Ni and contains metal veins. One curvilinear metal vein is metallic Fe-Ni grains in the interior appear unaltered; they lack limonite rims and are consistent with weathering stage Table 1. Modal abundances of major phases in Spade. a W1 (Wlotzka 1993). However, within 3– 4 mm of the Vol% Wt% Mean H (wt%) b meteorite surface, brown staining is much more extensive and Silicate 84.5 73.5 74.7 most metal and sulfide grains have been largely or completely 5.4 6.6 5.3 altered. Numerous subparallel 10–100 mm-thick veins and Metallic Fe-Ni 8.7 18.2 18.6 associated veinlets of limonite occur in these extensively Chromite 0.6 0.7 0.6 weathered regions, consistent with weathering stage W5. Limonite 0.8 0.9 – Total 100.0 99.9 100.0 Chondrite Group Nr of points 2190 2190 – Area (mm2) 200 200 – Modal analysis (Table 1) indicates that Spade contains aVol% percent was converted into wt% using the following specific gravities: 18.2 wt% metallic Fe-Ni; this is half way between the silicate, 3.3; troilite, 4.67; metallic Fe-Ni, 7.95; chromite, 4.7; limonite, normative mineralogical value of H chondrites (18.6 wt%; 4.28. bNormative mineralogy from Table 4.1 in Dodd (1981); total includes Table 4.1 in Dodd 1981) and the mean of H chondrite falls 0.6 wt% apatite and 0.2 wt% ilmenite. analyzed by wet chemistry (17.8 wt%; Jarosewich 1990).

Table 2. Compositions (wt%) of silicate and oxide phases in Spade. a Olivine Low-Ca pyx Ca pyx Plagioclase Chromite Nr of grains 9 7 10 24 10

SiO2 39.1 ± 0.4 55.7 ± 0.4 53.6 ± 0.3 64.8 ± 2.0 <0.04 TiO2 <0.04 0.27 ± 0.05 0.41 ± 0.06 – 1.9 ± 0.2 Al2O3 <0.04 0.36 ± 0.09 0.99 ± 0.10 22.6 ± 1.5 3.5 ± 0.7 Cr2O3 0.04 ± 0.01 0.57 ± 0.11 1.5 ± 0.12 n.d. 60.6 ± 0.7 V2O3 n.d. n.d. n.d. n.d. 0.67 ± 0.07 FeO 17.9 ± 0.3 10.7 ± 0.1 5.1 ± 0.2 0.55 ± 0.14 27.4 ± 1.1 MnO 0.52 ± 0.02 0.52 ± 0.05 0.29 ± 0.03 n.d. 0.91 ± 0.08 MgO 42.9 ± 0.2 30.2 ± 0.4 17.8 ± 0.2 <0.04 4.8 ± 0.7 CaO <0.04 1.8 ± 0.2 19.3 ± 0.2 3.3 ± 1.7 n.d. Na2O <0.04 0.08 ± 0.03 0.73 ± 0.07 9.4 ± 0.7 n.d. K2O n.d. <0.02 <0.02 0.86 ± 0.48 n.d. Total 100.5 100.2 99.8 101.5 99.8 End member Fa19.0 ± 0.3 Fs16.0 ± 0.1Wo3.5 ± 0.4 Fs8.4 ± 0.3Wo40.2 ± 0.5 Ab79.9 ± 5.7Or4.8 ± 2.7 an.d. = not determined. Spade: An H chondrite impact-melt breccia 1509

Table 3. Compositions (wt%) of opaque phases in Spade. a Modal analysis (2868 points) of the thin section indicates Kamacite Plessite Troilite Pent that Spade contains 1.8 vol% relict barred olivine (BO) and Nr of grains 28 9 18 1 porphyritic olivine-pyroxene (POP) chondrules. The relict Fe 92.8 ± 1.0 88.0 ± 3.5 63.8 ± 0.3 56.4 chondrules are highly recrystallized and range in size from Ni 6.9 ± 0.5 11.6 ± 3.3 <0.04 8.7 300 to 1000 mm. They exhibit extensive silicate darkening Co 0.46 ± 0.03 0.34 ± 0.08 <0.04<0.04 caused by the dispersion in the silicate phenocrysts of 0.2–4 S n.d. n.d. 35.7 ± 0.7 34.2 mm-size opaque blebs, mainly troilite. Some relict chondrules Total 100.2 99.9 99.5 99.3 are adjacent to small fine-grained regions and appear to be an.d. = not determined; pent = . overgrown by numerous 25 mm-size olivine grains that crystallized from the melt. The modal abundance of relict ~4 cm long and ~0.8 mm thick. Thin section observations chondrules in Spade is much lower than those in typical H5 show that Spade has a compact, recrystallized texture and H6 chondrites (i.e., 22.1 ± 6.7 vol% and 11.4 ± 3.6 vol%, composed of interlocking silicate (olivine, low-Ca pyroxene, respectively; Table 2 in Rubin et al. 2001). Ca pyroxene, plagioclase), metallic Fe-Ni (kamacite, , Spade also contains 1.8 vol% coarse ( ³200 mm-size) plessite), sulfide (troilite, pentlandite), and chromite grains. mafic silicate grains, most of which exhibit extensive silicate Although many of the olivine and pyroxene grains occur darkening (Figs. 2a, 2b, and 3). The modal abundance of these in relatively coarse clusters, throughout the meteorite, grains is much lower than those in typical H5 and H6 interstitial fine-grained regions exist (Fig. 1), consisting of (in chondrites (i.e., 9.0 ± 2.1 vol% and 9.8 ± 2.1 vol%, vol%): ~70% olivine, 15% plagioclase, 10% low-Ca pyroxene, respectively; Table 2 in Rubin et al. 2001). and 5% metallic Fe-Ni and sulfide. Grain sizes in these areas Plagioclase patches are typically 100–200 mm in size and are typically 10–25 mm. In some cases, the olivine grains are are irregular in shape. Many plagioclase grains in the thin adjacent to one another and are tightly packed; in others, they section and across the 8.5 cm-long slab are shaped like pincers are separated by 10–50 mm-size patches of plagioclase. These and wrap around (i.e., partially to completely enclose) adjacent regions seem similar to impact-melted regions in the Ramsdorf 5–20 mm-size olivine grains (Fig. 4). Some plagioclase grains impact-melt rock (e.g., Fig. 2c in Yamaguchi et al. 1999). have 2 or 3 sets of pincer-shaped appendages. Most plagioclase

Fig. 1. Fine-grained region consisting mainly of 10–25 mm-size grains of olivine, plagioclase, and low-Ca pyroxene. Metal and troilite grains (right of center) appear black. Transmitted light. 1510 A. E. Rubin and R. H. Jones

Fig. 2. Coarse olivine grain containing numerous small opaque blebs (mainly troilite with minor metallic Fe-Ni); most of the small opaque blebs are within curvilinear trails that transect the olivine grain: a) transmitted light view showing significant silicate darkening; b) reflected light view showing some of the small opaque blebs (white). The abundance of blebs appears lower in reflected light because only those blebs that are at the surface of the 30 mm-thick thin section are visible.

patches are adjacent to (or partly surround) metallic Fe-Ni and/ Ca pyroxene is also relatively homogeneous in or troilite grains; a few plagioclase patches are completely composition (8.4 ± 0.3 mol% Fs). Its CaO content of surrounded by mafic silicate grains. 19.3 wt% (40.2 ± 0.5 mol% Wo) puts it in the compositional Olivine and low-Ca pyroxene are relatively range of endiopside. homogeneous in composition (Fa 19.0 ± 0.3 and Fs16.0 ± 0.1, Plagioclase compositions are somewhat variable respectively; Table 2), indicative of equilibrium and typical of (Table2). The An component (An 15.3 ± 7.9) ranges from 4.6 to H4–6 chondrites. However, the CaO content of the low-Ca 28.8 mol%; the Or component (Or 4.8 ± 2.7) ranges from 1.4 to pyroxene (1.8 wt%; 3.5 ± 0.4 mol% Wo) is much higher than 12.0 mol%. Individual plagioclase grains in Spade are that in typical H chondrites (e.g., Wo <2.4 mol%; Scott et al. compositionally zoned; they possess relatively calcic, low- 1986). The apparent absence of polysynthetically twinned Na2O cores (Ab 68–71Or2.2–2.5) and relatively sodic and low-Ca pyroxene suggests that all of the low-Ca pyroxene in potassic rims (Ab 83–85Or6–8). Spade is orthorhombic. Chromite occurs as 25–200 mm-size grains adjacent to Spade: An H chondrite impact-melt breccia 1511

Fig. 3. Numerous small blebs of metallic Fe-Ni (white) inside mafic silicate grains (dark gray) causing significant silicate darkening in Spade. Back scattered electron image. The scale bar is 10 mm long.

Fig. 4. Pincer-shaped plagioclase patches (dark gray) partially to completely surrounding small mafic silicate grains (medium gray). The plagioclase has apparently melted and flowed following impact heating. Many of the plagioclase patches are adjacent to metallic Fe-Ni (white). Back scattered electron image. The scale bar is 10 mm long. 1512 A. E. Rubin and R. H. Jones

metallic Fe-Ni and troilite, as isolated grains completely silicate boundary in the Portales Valley impact-melt breccia surrounded by mafic silicate, and in chromite veinlets and (e.g., Ruzicka et al. 1999; Rubin et al. 2001). chromite-plagioclase assemblages (described below). The Many grains of metallic Fe-Ni in the thin section and mean composition of chromite outside the veinlets and across the slab consist of kamacite surrounding rounded to chromite-plagioclase assemblages in Spade is appreciably oblong patches of duplex plessite (Figs. 6a and 6b); the lower in Al 2O3 (3.5 wt%) and FeO (27.4 wt%) and higher in plessite constitutes 10–80 vol% of these metallic Fe-Ni grains. MgO (4.8 wt%) than that of equilibrated H chondrites (5.6– (As defined by Buchwald [1975], e.g., Fig. 129, duplex 7.3 wt% Al2O3; 29.4–32.0 wt% FeO; 2.0–3.2 wt% MgO; plessite consists of intricate intergrowths of 0.5– 2 mm-size Table 2 in Bunch et al. 1967; see also Fig. 191 of Brearley and kamacite and taenite particles.) The plessite regions in Spade Jones [1998]). metal are typically surrounded by a ~0.5–2 mm-thick rim of taenite. The duplex plessite consists of 1 × 5 mm- to 6 × 30 mm- Metallography size sparks and spindles of kamacite surrounded by irregularly shaped fine-grained plessite regions. A few metallic Fe-Ni Although the largest metallic Fe-Ni grains in Spade are grains are completely composed of duplex plessite. quasi-equant and ~1.2 mm in maximum dimension, such Some kamacite grains enclose patches of net plessite grains are rare. The mean size of metal grains is 170 ± 95 mm consisting of 2– 10 mm-diameter rounded kamacite blebs (n = 108), comparable to, or slightly larger than, those in flanking 5 × 40 mm-size kamacite shafts; some kamacite sparks typical H5 and H6 chondrites (120 ± 150 mm; Table 2 of and spindles also occur in these grains. (As defined by Rubin et al. 2001). Some metallic Fe-Ni grains consist Buchwald [1975], e.g., Fig. 124, net plessite consists of 1–10 exclusively of kamacite; many of these are adjacent to small mm-size islands of taenite within kamacite that contains patches of troilite. Most kamacite-troilite interfaces are numerous boundaries and subboundaries.) A few net plessite curved, suggestive of an igneous origin (Scott 2003, personal occurrences in Spade are coarser and contain quasi-equant 5– communication). None of the metal grains in Spade contains 15 mm-size cells and shafts of kamacite (Fig. 6c). In some . cases, a 3–4 mm-thick zone of very fine-grained plessite occurs Some kamacite grains are partly surrounded by relatively between the net plessite and the 0.5 mm-thick taenite rim. coarse (~130–150 mm-size) grains of phosphate, either Electron microprobe analysis of plessite regions yields a by itself or merrillite plus chlorapatite (Fig. 5). mean Ni content of 11.6 ± 3.3 wt% (Table 3) and a range of Some similarly sized grains of merrillite occur at the metal- 8.4–16.6 wt% Ni. This variability is due to electron beam

Fig. 5. Kamacite grain (white) rimmed by two phosphate phases. The innermost rim (medium gray) is merrillite; the outer rim (light gray) is chlorapatite. The entire assemblage is surrounded by silicate (dark gray to black). Back scattered electron image. The scale bar is 100 mm long. Spade: An H chondrite impact-melt breccia 1513

Sulfide

Troilite grains in Spade have a mean size of 75 ± 55 mm (n = 109), comparable to those in typical H5 (75 ± 70 mm) and H6 chondrites (80 ± 60 mm) (Table 2 in Rubin et al. 2001). Most troilite grains are adjacent to coarse grains of metallic Fe-Ni and form curved interfaces with the metal (Fig. 7). Electron microprobe analysis indicates that the troilite grains are pure FeS (Table 3); Ni and Co are below the limits of detection (~0.04 wt%). One rare grain of pentlandite with 8.7 wt% Ni was analyzed (Table 3).

Shock Features

Shock Stage Although most small olivine grains in fine-grained regions are unstrained and exhibit sharp optical extinction, many coarse olivine grains outside the fine-grained regions exhibit undulose extinction. This indicates that the Spade whole-rock is shock stage S2 (Stöffler et al. 1991). This low degree of shock stands in contrast to such shock features as extensive silicate darkening (Figs. 2 and 3) and the occurrence of chromite-plagioclase assemblages (see below).

Silicate Darkening Spade exhibits extensive silicate darkening caused by randomly distributed 0.5–4 mm-size blebs of metallic Fe-Ni, troilite, and chromite inside mafic silicate grains. Also present in the silicate grains throughout the meteorite are 20–50 mm- long curvilinear trails of opaque grains. Most trails consist of 0.5–4 mm-size blebs of troilite, a few trails are composed of blebs of metallic Fe-Ni and troilite, and some trails consist of 0.2–0.5 mm-size blebs of chromite. A few of the chromite trails (e.g., Fig. 8) contain 0.7 × 2 mm-size chromite needles and chromite blebs.

Rapidly Solidified Troilite-Metal Intergrowths Many of the troilite grains contain 2–15 vol% small (2–4 mm-size) blebs, cells, and irregularly shaped patches of Fig. 6. Grains of metallic Fe-Ni consisting of kamacite (light gray to metallic Fe-Ni (e.g., Fig. 7). They resemble the rapidly white) surrounding patches of plessite in nital-etched thin section: a) solidified troilite-metal mixtures that are present in many a large patch of duplex plessite (medium gray) consisting of small shocked OC (e.g., Scott 1982). Possibly, these mixtures sparks and spindles of kamacite. The plessite constitutes ~50 vol% of represent late-stage shock and do not date from the event that the metal grain; b) a patch of duplex plessite containing small sparks and spindles of kamacite; this plessite patch constitutes ~20 vol% of caused wide-spread melting in Spade. the metal grain in which it occurs; c) two plessite patches within a kamacite grain. Duplex plessite occurs at left. The patch of net plessite Chromite-Plagioclase Assemblages on the right is composed of rounded kamacite blebs and elongated Approximately 15 vol% of the chromite in Spade (thin kamacite shafts. A thick taenite rim (white; center) occurs at the upper section and slab) occurs in chromite-plagioclase assemblages left side of the net plessite patch. All images are in reflected light. (Fig. 9). Modal abundance determinations (2190 points) overlap of small kamacite and taenite grains in different indicate that Spade contains 0.5 vol% of these assemblages. proportions. The analyzed spots with higher Ni contents The assemblages range in size from ~50– 620 mm and in contain a higher proportion of taenite and have lower Co chromite/plagioclase modal ratio from 0.05–0.33. The largest contents than the lower-Ni spots (which contain a higher chromite-plagioclase assemblage observed in Spade is quasi- proportion of kamacite). equant and ~620 mm in diameter. It consists of ~20 vol% 1514 A. E. Rubin and R. H. Jones

Fig. 7. Troilite grain (medium gray) flanked by coarse metallic Fe-Ni grains (white) and containing small irregularly shaped metal patches (white; center). Similar fine-grained metal-troilite intergrowths with high troilite/metal modal-abundance ratios occur throughout the meteorite. Reflected light.

Fig. 8. Narrow chromite veinlet (light gray; center) transecting mafic silicate grains (medium gray) and occurring adjacent to melted plagioclase (dark gray; right). Reflected light. Spade: An H chondrite impact-melt breccia 1515

Fig. 9. Fine-grained chromite-plagioclase assemblages: a) melted, pincer-shaped plagioclase grain (dark gray) that contains several patches of very fine-grained chromite blebs (light gray). The assemblage has a high plagioclase/chromite modal abundance ratio; b) pincer-shaped plagioclase patches containing a few small chromite grains (center) and a few coarse chromite grains (left of center). Both images are in reflected light. chromite (occurring as 0.1–5 mm-size euhedral, subhedral, The finest-grained assemblages consist of 95 vol% rounded, and anhedral grains) and ~80 vol% plagioclase. plagioclase and 5 vol% chromite occurring as 1–2 mm-size Chromite grains are concentrated in the center of the angular and rounded grains (Figs. 9a and 9b). As noted by assemblage; the outer 200 mm-thick region contains relatively Rubin (2003), a positive correlation seems to exist between few chromite grains. Approximately 0.5 vol% of the chromite grain size and modal abundance in the chromite- assemblage consists of small troilite blebs occurring among plagioclase assemblages: finer-grained assemblages tend to the small chromite grains. A few rounded 4 mm-diameter contain smaller proportions of chromite. chromite grains are surrounded by 0.2 mm-thick troilite rims. Several small chromite-plagioclase assemblages have DISCUSSION textures that resemble one another. A typical example is the smallest (50 × 90 mm) assemblage, which contains 25 vol% Formation of Opaque Assemblages chromite, present in the plane of the section as 4, relatively large (10–22 mm-size) grains and approximately a dozen Unshocked to weakly shocked type-4–6 OC contain small (0.3–2 mm-size) rounded blebs and angular grains. opaque assemblages that consist of kamacite, taenite, and 1516 A. E. Rubin and R. H. Jones

troilite; no plessite (except zoneless plessite; see below) is An impact melting history for plagioclase in Spade is present (e.g., W ood 1967). In contrast, abundant plessite consistent with the irregular, pincer shapes of the grains occurs in those OC that have suffered appreciable degrees of (Fig.4). The grains appear to have melted and flowed; during shock. These rocks include L5 Kingfisher (Taylor and this process, they partially to completely wrapped themselves Heymann 1970), H5 Sweetwater (Taylor and Heymann 1971), around adjacent grains of mafic silicate. We have observed H6 Guareña (Willis and Goldstein 1983), and LL6 MIL 99301 similar pincer-shaped plagioclase patches in the very strongly (Rubin 2002). Although zoneless plessite forms in chondrites shocked (shock stage S6; Bennett and McSween 1996a) during slow cooling (Sears and Axon 1975; Reisener and Kingfisher L5 chondrite. Goldstein Forthcoming), other forms of plessite are produced when highly localized stress concentrations resulting from an Low-Ca Pyroxene heat kamacite-taenite intergrowths above the Fe- The CaO content of low-Ca pyroxene in Spade (3.5 ± 0.4 Ni solvus. Moderately rapid cooling could result in the mol% Wo) exceeds that of unshocked H chondrites. The Wo formation of plessite. Alternatively, quenching could cause content of low-Ca pyroxene grains in OC does not vary martensite formation, and martensite could unmix to form significantly among the H, L, and LL groups but does plessite during annealing. The fact that most metallic Fe-Ni correlate weakly with petrologic type (Scott et al. 1986; grains in Spade contain plessite (Fig. 6) indicates that shock Brearley and Jones 1998). Nevertheless, the Wo contents of heating of the rock was pervasive. low-Ca pyroxene in H4–6 chondrites do not generally exceed Likely, during annealing, a large fraction of P diffused 2 mol%. Although the ALH A81187 has a out of the metal and was oxidized to form Ca phosphate relatively high W o content (Fs 6.9Wo3.2; Table 20 in (Murrell and Burnett 1983) at the metal-silicate interface. Mittlefehldt et al. 1998), its Fs content is much lower than that Calcium was probably derived from plagioclase and of Spade (i.e., Fs 16.0 ± 0.1), and it is not directly comparable. pyroxene. Concomitant reduction of FeO to metallic Fe could Elevated Wo contents (~3 mol%) in low-Ca pyroxene have supplied the oxygen. If this is the case, then some of the were observed in material that equilibrated with partially small metal blebs within mafic silicate grains in Spade molten chondritic material at 1200°C within a few days possibly formed by reduction of FeO and not by shock (Feldstein et al. 2001). In general, the CaO contents in low-Ca mobilization of metallic Fe-Ni. pyroxene correlate positively with the temperature of formation of the grains (whether these grains formed by Implications of Silicate Mineralogy crystallization from a melt, equilibration with a melt at high temperatures, or crystallization during thermal Plagioclase metamorphism) (e.g., Atlas 1952; Schairer and Boyd 1957; Equilibrated OC tend to contain plagioclase that is Dobretsov 1968). relatively uniform in composition, e.g., Kernouvé (H6) Pyroxene thermometry can yield approximate plagioclase has a mean composition of An 11.4 ± 0.8 and a range equilibration temperatures for Spade. Because the Mg-Fe of 9.3 to 12.7 An (n = 12); Ogi (H6) plagioclase has a mean thermometer of Kretz (1982) gives temperatures that are too composition of An 12.5 ± 0.4 and a range of 12.1 to 12.9 An (n = high for some pyroxene grains that were equilibrated 7) (Rubin 1992). In contrast, Spade plagioclase has a experimentally (Lindsley 1983), the Lindsley thermometer is moderately large range in composition (An 4.6–28.8; n = 24), used here to estimate equilibration temperature. The Wo and that more closely resembles the ranges observed in the Rose Fs mole fractions of Ca pyroxene (0.385 and 0.140) and low- City H chondrite impact-melt breccia (An 8.5–23.9; n = 11) and Ca pyroxene (0.035 and 0.160) were determined from the the moderately shocked (shock stage S4) Paragould LL5 data in Table 2 following the projection scheme of Lindsley chondrite (An 8.7–44.5; n = 16) (Rubin 1992). (1983) and Lindsley and Andersen (1983). Plotting these data Shock remelting may account for the compositional on the 1 atm diagram (Fig. 6 in Lindsley and Andersen 1983) zoning of individual plagioclase grains in Spade. These yields equilibration temperatures of 1100°C for Ca pyroxene grains have cores with relatively high CaO and low K 2O and 1200°C for low-Ca pyroxene. (The Mg-Fe thermometer (Ab68–71Or2.2–2.5) and rims with relatively high Na 2O and of Kretz [1982] yielded a much higher equilibration K2O (Ab83–85Or6–8), consistent with an igneous zoning temperature, i.e., 1330°C.) The inferred equilibration pattern. Coarse plagioclase grains in equilibrated OC tend to temperatures derived from the Lindsley method are the same have uniform (i.e., unzoned) CaO contents; they also tend to as (or somewhat higher than) that of the OC solidus have nonuniform K 2O contents that vary by a factor of ~2 temperature (~1100°C; Jurewicz et al. 1995), consistent with within individual grains (Van Schmus and Ribbe 1968). Spade having been at least partly melted. (However, the K 2O variations in plagioclase from A high temperature origin is, thus, likely for Spade and equilibrated OC do not correlate with crystal geometry, i.e., for rocks that contain low-Ca pyroxene compositions close to no evidence exists for systematic compositional zoning of those of Spade (Fs 16.0Wo3.5); the latter include some K2O in these plagioclase grains.) orthopyroxene-bearing (e.g., Y -74130; Fs 17.8Wo4.2; Spade: An H chondrite impact-melt breccia 1517

Table 23 of Mittlefehldt et al. 1998) and silicate inclusions in eutectic temperature of 988° C and mobilization of the the Watson IIE iron (Fs 17.3Wo3.7; Olsen et al. 1994). engendered melt. The curvilinear trails of monomineralic chromite formed Relict Chondrules and Coarse Mafic Silicate Grains from chromite-plagioclase melts that were injected into fractures in adjacent silicate grains (Rubin 2003); the The low modal abundance of relict chondrules in Spade chromite crystallized and the residual plagioclase-rich melt (1.8 vol%) compared to average H6 chondrites (11.4 vol%) continued to flow, eventually pooling to form irregularly suggests that >80% of the chondrules in Spade were shaped plagioclase patches and melt pockets. The extensive obliterated. This is a lower limit. If Spade had been a normal silicate darkening in Spade (among both coarse and fine mafic H3 chondrite with 65–75 vol% chondrules (e.g., Table 1 in silicate grains) indicates that essentially all of the silicate Rubin 2000) before the impact event, then >95% of the grains in the whole-rock were invaded by metal-sulfide melts. chondrules would have been obliterated. Those silicate grains that did not melt retained the opaque obliteration can be caused by brecciation, metamorphic melt trails; they , thus, appear dark when viewed recrystallization, and impact melting. In view of the other microscopically in transmitted light. evidence for shock (e.g., silicate darkening, melted plagioclase, chromite veinlets, chromite-plagioclase Petrogenesis assemblages) and the inferred equilibration temperatures for pyroxene that are at or above the OC solidus, impact melting We conclude that Spade is an impact-melt breccia that is likely to be primarily responsible for chondrule obliteration has undergone appreciable melting of its chondritic precursor in Spade. material. Because melting was extensive, the petrologic type Chondrule loss by impact melting has been inferred of Spade before the impact event is unclear. However, the low previously for the Portales Valley H chondrite (Rubin et al. Al2O3 content of the matrix chromite grains (3.5 ± 0.7 wt%) 2001) and the EH (Rubin and Scott compared to average H4–6 chondrites (5.6–7.3 wt% Al 2O3; 1997). In both of these , the relict chondrules Bunch et al. 1967) may indicate that Spade’s progenitor was served as nucleation sites for mafic silicates growing from the H3; chromite with low Al 2O3 occurs in some type-3 OC such surrounding impact melt. The same is true for Spade. As as H3.6 Prairie Dog Creek (2.7 wt% Al 2O3) and H/L3.9 described above, some relict chondrules appear to be Bremervörde (0.2 wt% Al 2O3) (Bunch et al. 1967). Although, overgrown by numerous 25 mm-size olivine grains. conceivably, Spade was metamorphosed and partly melted by Isolated 200 mm-size mafic silicate grains in OC are the decay of 26Al, the occurrence of chromite-plagioclase probably derived from broken chondrules (e.g., Nelson and assemblages and chromite veinlets suggests that impact- Rubin 2002). Spade contains far fewer of these grains melting was responsible for Spade’s temperature excursion. (1.8vol%) than average H5 and H6 chondrites (9.0 ± The shock event caused Spade to reach shock stage S6. 2.1 vol% and 9.8 ± 2.1 vol%, respectively; Table 2 in Rubin et Relict silicate grains became infused with metal-sulfide and al. 2001). The low abundance of coarse mafic silicate grains chromite-rich melts, resulting in extensive silicate darkening in Spade most likely indicates that impact-melting destroyed and the production of chromite veinlets. Chromite-plagioclase ³80% of the pre-existing coarse grains. assemblages also formed during the shock event. Plagioclase was melted and it flowed around nearby mafic silicate grains, Chromite-Plagioclase Assemblages partly enclosing some of them, resulting in the characteristic pincer shape of Spade plagioclase grains. Low-Ca pyroxene Chromite-plagioclase assemblages occur in nearly every crystallizing from the melt (or equilibrating with the melt at shock stage S3–S6 OC and are a useful petrographic indicator high temperatures) acquired relatively high amounts of CaO. of shock (Rubin 2003, Forthcoming). Plagioclase has a low Metallic Fe-Ni cooled rapidly below the Fe-Ni solvus and impedance to shock compression; that heat from shock- transformed into martensite. Subsequent reheating of the rock melted plagioclase caused adjacent chromite grains to melt (discussed below) may have been responsible for the seems plausible. Chromite crystallized from this melt. The unmixing of the martensite and the development of plessite. residual melt quenched and chromite-plagioclase Coarse grains of phosphate associated with the metallic Fe-Ni assemblages were produced. probably also grew during post-shock annealing.

Silicate Darkening and Chromite Veinlets Burial and Exhumation

Silicate darkening is caused by the dispersion of small There are ambiguities in the shock stage assignment of blebs of metal and/or troilite inside and around mafic silicate Spade. High shock stages (S3–S6) are indicated by the grains (Rubin 1992). This phenomenon results mainly from extensive silicate darkening (Rubin 1992) and the occurrence impact heating of metallic Fe-Ni and troilite grains above the of chromite-plagioclase assemblages and chromite veinlets 1518 A. E. Rubin and R. H. Jones

(Rubin 2002, 2003). The impact-melted characteristics of features in olivine, obliterate mosaicism, and leave the olivine Spade indicate shock stage S6 (corresponding to 45–60 GPa; grains with undulose extinction. More probably, heating Stöffler et al. 1991; Schmitt et al. 1994; Schmitt and Stöffler would have annealed Spade to S1 levels. During such an 1995). If the rock had been shocked to S6 levels and left annealing episode, the olivine crystal lattices would have otherwise undisturbed, the olivine grains would have healed; strain would have been reduced, and the olivine grains exhibited planar fractures, planar deformation features and would have developed sharp optical extinction. pronounced mosaic extinction (Stö ffler et al. 1991). Also, If we assume that Spade was annealed to S1 levels, then likely, low-Ca clinopyroxene would have formed: the occurrence of undulose extinction in its olivine grains Hornemann and Müller (1971) and Stöffler et al. (1991) indicates that the rock was (very weakly) shocked again after found that clinopyroxene lamellae parallel to (100) form annealing. This may have occurred during exhumation from within orthopyroxene grains at shock pressures as low as the crater or from the ejecta blanket, possibly by the same ~5 GPa. collisional event that launched Spade off the H chondrite However, olivine in Spade is characteristic of shock stage parent . This scenario identifies Spade as an H S2; it exhibits undulose extinction and lacks planar fractures chondrite that experienced impact melting, post-shock and planar deformation features. In addition, low-Ca annealing, and post-annealing shock. clinopyroxene appears to be absent. These features suggest Because Spade is an impact-melt breccia, no doubt exists that Spade may have been shocked to stage S6 and then that it was once severely shocked, most likely to shock stage annealed. Annealing temperatures must have exceeded 630°C S6. The lack of mosaic extinction of its relict olivine (i.e., (the temperature at which low-Ca clinopyroxene transforms grains that did not crystallize from the impact melt) indicates into orthopyroxene; Boyd and England 1965; Grover 1972). that Spade was annealed, thus, firmly establishing the case for The annealing temperatures experienced by normal type-5 post-shock annealing. Likely, significant amounts of material and -6 OC (700–750° C and 820–930° C, respectively; Dodd on OC asteroids experienced successive episodes of thermal 1981; Olsen and Bunch 1984) are below that inferred for metamorphism, shock metamorphism, and post-shock Spade from pyroxene thermometry (1100–1200° C). annealing. Some of this material may have been shocked The absence of schreibersite associated with metallic Fe- again. Possibly, some OC material experienced multiple Ni in Spade suggests that cooling was slow enough for P to episodes of shock and annealing. This inference is consistent have been oxidized and reacquired by phosphate. with the shock history inferred from 40Ar-39Ar studies of Although shock-induced features in olivine seem to have some impact-melt-bearing OC (Kunz et al. 1997). been annealed, Spade retains heterogeneous feldspar compositions. These observations allow us to place Acknowledgments–We thank J. T. W asson (UCLA) for constraints on the thermal history of the rock. Bauer (1979) discussions and Mr. J. Talbert and Dr. C. Barnes (Texas Tech showed that it is possible to anneal shock-induced University) for providing the sample of Spade. We appreciate microfractures in olivine in short times (20 minutes) in the the detailed and helpful reviews by E. R. D. Scott and A. temperature range 700–900º C. However, longer heating times Yamaguchi and the comments of Associate Editor H. tend to produce recrystallized grains. The Na, K Nagahara. This work was supported in part by NASA Grant interdiffusion rates in alkali feldspar are on the order of 10 -18 NAG5–4766 (A. E. Rubin) and NASA Grant NAG5– 9463 to 10-19 m2 s-1 at about 800ºC (Hokanson and Yund 1986). (R. H. Jones; J. J. Papike, P. I.). This means that Na, K interdiffusion on a scale of several mm, sufficient to result in homogenization of the feldspar, would Editorial Handling—Dr. Hiroko Nagahara take place in about 1 yr. These 2 indicators suggest that post-shock heating times REFERENCES were fairly short. This is consistent with the annealing period

expected for rocks in close proximity to an impact crater, Atlas L. 1952. The polymorphism of MgSiO 3 and solid state perhaps in a melted region beneath the crater floor or near hot equilibria in the system MgSiO 3-CaMgSi2O6. Journal of debris within ejecta blankets at the asteroid surface. Global Geology 60:125–147. heating caused by the decay of 26Al should occur over Bauer J. F. 1979. Experimental shock metamorphism of mono- and polycrystalline olivine: A comparative study. Proceedings, 10th significantly longer time scales, i.e., millions of years (e.g., Lunar and Planetary Science Conference. pp. 2573–2596. Bennett and McSween 1996b; Keil 2000). Bennett M. E. and McSween H. Y. 1996a. Shock features in iron- Whole-rock annealing is also probably responsible for nickel metal and troilite of L-group ordinary chondrites. the unmixing of the martensite and the formation of plessite Meteoritics & Planetary Science 31:255–264. (Fig. 6). Cobalt and Fe partitioned into kamacite, and Ni Bennett M. E. and McSween H. Y. 1996b. Revised model calculations for the thermal histories of parent partitioned into taenite. bodies. Meteoritics & Planetary Science 31:783–792. Annealing seems unlikely to have been sufficiently Bogard D. D., Garrison D. H., Norman M., Scott E. R. D. and Keil precise to remove planar fractures and planar deformation K. 1995. 39Ar-40Ar age and petrology of Chico: Large-scale Spade: An H chondrite impact-melt breccia 1519

impact melting on the L chondrite . Geochimica et Nelson V. E. and Rubin A. E. 2002. Size-frequency distributions of Cosmochimica Acta 59:1383–1400. chondrules and chondrule fragments in LL3 chondrites: Boyd F. R. and England J. L. 1965. The rhombic enstatite- Implications for parent-body fragmentation of chondrules. clinoenstatite inversion. In Annual report of the Director, Meteoritics & Planetary Science 37:1361–1376. Geophysical laboratory. Washington D.C.: Carnegie Institution Olsen E. J. and Bunch T. E. 1984. Equilibration temperatures of the of Washington. pp. 117–120. ordinary chondrites: A new evaluation. Geochimica et Brearley A. J. and Jones R. H. 1998. Chondritic meteorites. In Cosmochimica Acta 48:1363–1365. Planetary materials, edited by Papike J. J. Washington D.C.: Olsen E. J., Davis A., Clarke R. S., Schultz L., Weber H., Clayton R. Mineralogical Soecity of Amercia. pp. 3-1–3-398. N., Mayeda T., Jarosewich E., Sylvester P., Grossman L., Wang Dobretsov N. L. 1968. Paragenetic types and compositions of M. S., Lipschutz M. E., Steele I. M., and Schwade J. 1994. metamorphic . Journal of Petrology 9:358–377. Watson: A new link in the IIE iron chain. Meteoritics 29:200– Dodd R. T. 1981. Meteorites—A petrologic-chemical synthesis. New 213. York: Cambridge University Press. 368 p. Reisener R. J. and Goldstein J. I. Forthcoming. The formation of Feldstein S. N., Jones R. H. and Papike J. J. 2001. Partial melting zoned and unzoned metal particles in low-shock types 4–6 experiments on the Leedey L6 chondrite: The textural controls of ordinary chondrites. Meteoritics & Planetary Science . melting. Meteoritics & Planetary Science 36:1421–1441. Rubin A. E. 1990. Kamacite and olivine in ordinary chondrites: Fodor R. V. and Keil K. 1976. Carbonaceous and non-carbonaceous Intergroup and intragroup relationships. Geochimica et lithic fragments in the Plainview, Texas chondrite: Origin and Cosmochimica Acta 54:1217–1232. history. Geochimica et Cosmochimica Acta 40:177–189. Rubin A. E. 1992. A shock-metamorphic model for silicate darkening Grover J. E. 1972. The stability of low-clinoenstatite in the system and compositionally variable plagioclase in CK and ordinary Mg2Si2O6-CaMgSi2O6 (abstract). Transactions of the American chondrites. Geochimica et Cosmochimica Acta 56:1705–1714. Geophysical Union 53:539. Rubin A. E. 2000. Petrologic, geochemical, and experimental Hokanson S. A. and Yund R. A. 1986. Comparison of alkali constraints on models of chondrule formation. Earth-Science interdiffusion rates for cryptoperthites. American Mineralogist Reviews 50:3–27. 71:1409–1414. Rubin A. E. 2002. Post-shock annealing of Miller Range 99301: Hornemann U. and Müller W. F. 1971. Shock-induced deformation Implications for impact heating of ordinary chondrites. twins in clinopyroxene. Neues Jahrbuch fuer Mineralogie 6: Geochimica et Cosmochimica Acta 66:3327–3337. 247–256. Rubin A. E. 2003. Chromite-plagioclase assemblages as a new shock Jarosewich E. 1990. Chemical analyses of meteorites: A compilation indicator; Implications for the shock and thermal histories of of stony and analyses. Meteoritics 25:323–337. ordinary chondrites. Geochimica et Cosmochimica Acta 67: Jurewicz A. J. G., Mittlefehldt D. W., and Jones J. H. 1995. 2695–2709. Experimental partial melting of the St. Severin (LL) and Lost Rubin A. E. Forthcoming. Postshock annealing and postannealing City (H) chondrites. Geochimica et Cosmochimica Acta 59:391– shock in equilibrated ordinary chondrites: Implications for the 408. thermal and shock histories of chondritic asteroids. Geochimica Keil K. 2000. Thermal alteration of asteroids: Evidence from et Cosmochimica Acta . meteorites. Planetary and Space Science 48:887–903. Rubin A. E. and Scott E. R. D. 1997. Abee and related EH chondrite Kretz R. 1982. Transfer and exchange equilibria in a portion of the impact-melt breccias. Geochimica et Cosmochimica Acta 61: pyroxene quadrilateral as deduced from natural and experimental 425–435. data. Geochimica et Cosmochimica Acta 46:411–421. Rubin A. E., Rehfeldt A., Peterson E., Keil K., and Jarosewich E. Kring D. A., Swindle T. D., Britt D. T., and Grier J. A. 1996. Cat 1983. Fragmental breccias and the collisional evolution of Mountain: A meteoritic sample of an impact-melted asteroid ordinary chondrite parent bodies. Meteoritics 18:179–196. regolith. Journal of Geophysical Research 101:29353–29371. Rubin A. E., Ulff-Møller F., Wasson J. T., and Carlson W. D. 2001. Kunz J., Falter M., and Jessberger E. K. 1997. Shocked meteorites: The Portales Valley meteorite breccia: Evidence for impact- Argon-40-argon-39 evidence for multiple impacts. Meteoritics & induced melting and metamorphism of an ordinary chondrite. Planetary Science 32:647–670. Geochimica et Cosmochimica Acta 65:323–342. Lindsley D. H. 1983. Pyroxene thermometry. American Mineralogist Ruzicka A., Snyder G. A., Prinz M., and Taylor L. A. 1999. Portales 68:477–493. Valley: A new metal-phosphate-rich meteorite with affinities to Lindsley D. H. and Andersen D. J. 1983. A two-pyroxene Netschaëvo and H-group chondrites (abstract #1645). 30th Lunar thermometer. Proceedings, 13th Lunar and Planetary Science and Planetary Science Conference. Conference. pp. A887–A906. Schairer J. F. and Boyd F. R. 1957. Pyroxenes, the join MgSiO 3- Mason B. and Wiik H. B. 1966. The composition of the Bath, CaMgSi2O6. Annual report of the Director, Geophysical Frankfort, Kakangari, Rose City, and Tadjera meteorites. laboratory. Washington D.C.: Carnegie Institution of American Museum Novitates 2272:1–24. Washington. pp. 223. Mittlefehldt D. W. and Lindstrom M. M. 2001. Petrology and Schmitt R. T., Deutsch A. and Stöffler D. 1994. Shock recovery geochemistry of Patuxent Range 91501, a clast-poor impact melt experiments with the H6 chondrite Kernouvé at preshock from the L chondrite parent body and Lewis Cliff 88663, an L7 temperatures of 293 and 920 K (abstract). Meteoritics 29:529– chondrite. Meteoritics & Planetary Science 36:439–457. 530. Mittlefehldt D. W., McCoy T. J., Goodrich C. A., and Kracher A. Schmitt R. T. and Stöffler D. 1995. Classification of chondrites 1998. Non-chondritic meteorites from asteroidal bodies. In (abstract). Meteoritics 30:574–575. Planetary materials, edited by Papike J. J. Washington D.C.: Scott E. R. D. 1982. Origin of rapidly solidified metal-troilite grains Mineralogical Society of America. pp. 4:1–4:195. in chondrites and iron meteorites. Geochimica et Cosmochimica Murrell M. T. and Burnett D. S. 1983. The behavior of actinides, Acta 46:813–823. phosphorus, and rare earth elements during chondrite Scott E. R. D., Taylor G. J., and Keil K. 1986. Accretion, metamorphism. Geochimica et Cosmochimica Acta 47:1999– metamorphism, and brecciation of ordinary chondrites: Evidence 2014. from petrologic studies of meteorites from Roosevelt County, 1520 A. E. Rubin and R. H. Jones

New Mexico. Proceedings, 17th Lunar and Planetary Science structural state of feldspar from chondritic meteorites. Conference. pp. E115–E123. Geochimica et Cosmochimica Acta 32:1327–1342. Sears D. W. and Axon H. J. 1975. The metal content of common Wilkening L. L. 1978. Tysnes Island: An unusual clast composed of chondrites (abstract). Meteoritics 10:486–487. solidified, immiscible, Fe-FeS and silicate melt. Meteoritics 13: Stöffler D., Keil K., and Scott E. R. D. 1991. Shock metamorphism 1–9. of ordinary chondrites. Geochimica et Cosmochimica Acta 55: Willis J. and Goldstein J. I. 1983. A three dimension study of metal 3845–3867. grains in equilibrated, ordinary chondrites. Proceedings, 14th Takeda H., Huston T. J., and Lipschutz M. E. 1984. On the Lunar and Planetary Science Conference. pp. B287–B292. chondrite- transition: Mineralogy and chemistry of Wlotzka F. 1993. A weathering scale for the ordinary chondrites Yamato-74160 (LL7). Earth and Planetary Science Letters 71: (abstract). Meteoritics 28:460. 329–339. Wood J. A. 1967. Chondrites: Their metallic minerals, thermal Taylor G. J. and Heymann D. 1970. Electron microprobe study of histories, and parent planets. Icarus 6:1–49. metal particles in the Kingfisher meteorite. G eochimica et Yamaguchi A., Scott E. R. D., and Keil K. 1998. Origin of unusual Cosmochimica Acta 34:677–687. impact melt rocks, Yamato-790964 and -790143 (LL chondrites). Taylor G. J. and Heymann D. 1971. Postshock thermal histories of Antarctic Meteorite Research 11:18–31. reheated chondrites. Journal of Geophysical Research 76:1879– Yamaguchi A., Scott E. R. D., and Keil K. 1999. Origin of a unique 1893. impact-melt rock—The L chondrite Ramsdorf. Meteoritics & Van Schmus W. R. and Ribbe P. H. 1968. The composition and Planetary Science 34:49–59.